REPORT STATUS DECLARATION FORMeprints.utar.edu.my/2466/1/CN-2017-1400454.pdf · Title: Aloha-Based Radio-Frequency Identification (RFID) System With Early Frame Adjustment Academic

UNIVERSITI TUNKU ABDUL RAHMAN

REPORT STATUS DECLARATION FORM

Title: Aloha-Based Radio-Frequency Identification (RFID) System With Early Frame

Adjustment

Academic Session: Jan 2017

I, LEE KHAI YI declare that I allow this Final Year Project Report to be kept in

Universiti Tunku Abdul Rahman Library subject to the regulations as follows:

1. The dissertation is a property of the Library.

2. The Library is allowed to make copies of this dissertation for academic purposes.

Verified by,

_________________________ _________________________

(Author’s signature) (Supervisor’s signature)

Address:

29, Jalan Besar

11000 Balik Pulau,

Pulau Pinang. __________________________

Dr Robithoh Annur

Date: _____________________ Date:____________________

ALOHA-BASED RADIO-FREQUENCY IDENTIFICATION (RFID) SYSTEM

WITH EARLY FRAME ADJUSTMENT

BY

LEE KHAI YI

TITLE PAGE

A REPORT

SUBMITTED TO

Universiti Tunku Abdul Rahman

In partial fulfillment of the requirement

for the degree of

BACHELOR OF INFORMATION AND COMMUNICATION TECHNOLOGY (HONS)

COMMUNICATION AND NETWORKING

Faculty of Information and Communication Technology

(Perak Campus)

JAN 2017

ii

DECLARATION OF ORIGINALITY

I declare that this report entitled “ALOHA-BASED RADIO-FREQUENCY

IDENTIFICATION (RFID) SYSTEM WITH EARLY FRAME ADJUSTMENT”

is my own work except as cited in the references. The report has not been accepted for

any degree and is not being submitted concurrently in candidature for any degree or

other award.

Signature : _________________________

Name : _________________________

Date : _________________________

BIT (Hons) Communications and Networking

Faculty of Information and Communication Technology (Perak Campus), UTAR. iii

ACKNOWLEDGEMENTS

Firstly, I would like to take this opportunity to present my gratitude to my

supervisor, Dr. Robithoh Annur who has given me the chance to involve in this

project and enables me to have the opportunity to study and explore the anti-collision

algorithms which helps to enhance RFID tag identification process. A faithful thanks

for all the assistance and guidance given throughout this project.

Next, I would like to thank my academic advisor, Miss Wong See Wan for

giving me useful advices and suggestions whenever I face any difficulties in my study.

Besides, I would also like to express my sincere appreciation to my family and loved

one who has always been pillars of strength for me, giving me unconditional support

and encouragement when any challenges or obstacles arise. A million thanks to them.

Last but not least, I would like to express my appreciation to my peers who

willing to spend their time providing me ideas and solutions whenever I face any

challenges or technical issues during project development phase.


Faculty of Information and Communication Technology (Perak Campus), UTAR. iv

ABSTRACT

Radio-Frequency Identification (RFID) is a technology which utilised

electromagnetic fields for fast object tracking and identification. The beauty of this

technology such as fast data reading without line of sight (LOS), large memory, long

service life and strong penetrability has made RFID becomes a technology employed

in different applications and commercial sectors to automate mundane tasks.

However, there are several drawbacks in RFID system has been discovered.

One of the disadvantages of using passive RFID system is it will lead to tag collision

when a reader is receiving signal sent from two or more tags at the same time.

Consequently, these signals could not differentiate by the reader and the hence the tag

information also could not receive correctly. Therefore, anti-collision algorithms that

help to prevent tags collision are needed.

There are two main types of RFID anti-collision algorithm which are Binary

Tree and ALOHA-based. In this project, we are mainly focus on ALOHA-based anti-

collision algorithms and we are going to study Frame Slotted ALOHA (FSA) and

Dynamic frame slotted ALOHA (DFSA) respectively. In most cases, DFSA is always

adopted to resolve tag collision in RFID system as it could provide dynamic frame

length that fits the collision situation during identification process. FSA is not

preferable due to its static initial frame size that could lead to very high number of

collision in the worst case.

However, the significant drawback of using DFSA is it has to predict the

number of tags correctly in order to offer an optimal frame size. This is the most

challenging task in DFSA. Thus, this project is going to propose a new timing concept

which would enhance the tag identification process and mitigate RFID tag collision

problem by utilising Manchester Coding, a bit-tracking technology in DFSA that

allows RFID reader to recognise the location of collision bits within a time slot.

Besides, Gen2 standard was applied in this project for the purpose of obtaining the

slot duration during tag identification process and the tag identification rate of FSA,

DFSA and proposed approach under different given scenarios.


Faculty of Information and Communication Technology (Perak Campus), UTAR. v

TABLE OF CONTENTS

TITLE PAGE ................................................................................................................. i

DECLARATION OF ORIGINALITY ....................................................................... ii

ACKNOWLEDGEMENTS ........................................................................................ iii

ABSTRACT .................................................................................................................. iv

TABLE OF CONTENTS ............................................................................................. v

LISTS OF TABLES .................................................................................................... vii

LISTS OF FIGURES ................................................................................................. viii

LIST OF ABBREVIATIONS ...................................................................................... x

CHAPTER 1: INTRODUCTION ................................................................................ 1

1.1 Problem Statement and Motivation ....................................................................... 1

1.2 Project Background ............................................................................................... 2

1.3 Objectives .............................................................................................................. 4

1.4 Proposed Approach ............................................................................................... 5

1.5 Report Organisation .............................................................................................. 7

CHAPTER 2: LITERATURE REVIEW ................................................................... 8

2.1 Review of technologies ......................................................................................... 8

2.1.1 RFID system ................................................................................................... 8

2.1.2 Generation 2 (Gen2) technology .................................................................... 9

2.1.3 Manchester Coding ....................................................................................... 10

2.2 Review of -based anti-collision algorithms ......................................................... 11

2.2.1 Pure ALOHA ................................................................................................ 11

2.2.2 Slotted ALOHA ............................................................................................ 11

2.2.3 Frame Slotted ALOHA (FSA) ...................................................................... 11

2.2.4 Dynamic Slotted ALOHA (DFSA) .............................................................. 12

2.2.5 Summary of ALOHA-based anti-collision algorithms ................................. 12

2.3 Review of existing tag estimation algorithms ..................................................... 13

2.3.1 Lowbound algorithm .................................................................................... 13

2.3.2 Schoute algorithm ......................................................................................... 13

2.3.3 Improved Linearized Combinatorial Model (ILCM) ................................... 13

2.3.5 Summary of existing tag estimation algorithms ........................................... 14

2.4 Review of existing improved ALOHA-based anti-collision algorithms ............. 15


Faculty of Information and Communication Technology (Perak Campus), UTAR. vi

2.4.1 Fitted Dynamic Framed Slotted ALOHA Anti-Collision Algorithm in RFID

Systems ......................................................................................................... 15

2.4.2 An Efficient and Easy-to-implement Tag Identification Algorithm for UHF

RFID Systems............................................................................................... 16

2.4.3 A Dynamic Framed Slotted ALOHA Anti-collision Algorithm Based on

Tag-Grouping for RFID Systems ................................................................. 18

2.4.4 Summary of existing improved ALOHA-based anti-collision algorithms ... 19

CHAPTER 3: SYSTEM DESIGN ............................................................................. 20

3.1 System flow ......................................................................................................... 20

Step 1: Frame size initialisation............................................................................. 21

Step 2: Tag distribution ......................................................................................... 21

Step 3: Slot reservation code generation ............................................................... 22

Step 4: Identify success, collision and empty slots ............................................... 22

Step 5: Tag estimation ........................................................................................... 25

Step 6: Gen2 timing implementation ..................................................................... 28

CHAPTER 4: IMPLEMENTATION AND ANALYSIS ........................................ 32

4.1 Design Specifications .......................................................................................... 32

4.1.1. Hardware ..................................................................................................... 32

4.1.2. Software ....................................................................................................... 32

4.2 Implementation of FSA and DFSA ..................................................................... 33

4.3 Results and discussion ......................................................................................... 36

CHAPTER 5: CONCLUSION................................................................................... 45

5.1 Project Review .................................................................................................... 45

5.2 Discussion ........................................................................................................... 46

5.3 Contributions ....................................................................................................... 47

5.4 Future works ........................................................................................................ 48

BIBLIOGRAPHY ....................................................................................................... 49

APPENDIX A - Weekly Report ............................................................................... A-1

APPENDIX B - Poster .............................................................................................. B-1

APPENDIX C - Turnitin Similarity Report ........................................................... C-1


Faculty of Information and Communication Technology (Perak Campus), UTAR. vii

LISTS OF TABLES

TABLE NUMBER TITLE PAGE

Table 2.1 Summary of ALOHA-based anti-collision

algorithms

12

Table 2.2 Summary of existing tag estimation algorithms 14

Table 2.3 Summary of existing improved ALOHA-based

anti-collision algorithms

19

Table 3.1 Tari, DR and BLF and their value ranges 30

Table 3.2 Equations for Rbl, PRT, Tpri, TRCal and RTCal

and Value Range

30

Table 3.3 Equations for TQuery, TACK, TQrep, T1, T2 and T3

and Value Range

31

Table 3.4 Equations for TRN16 and TEPC 31

Table 3.5 Equations for TS, TC and TE 31

Table 4.1 Gen2 parameters used in Scenario1, 2 and 3 40


Faculty of Information and Communication Technology (Perak Campus), UTAR. viii

LISTS OF FIGURES

FIGURE NUMBER TITLE PAGE

Figure 1.1 Diagram of RFID system collision types 2

Figure 1.2 Example of FSA tag identification process 3

Figure 1.3 Proposed Approach Flow Diagram 5

Figure 2.1 Diagram of how a RFID system works 8

Figure 2.2 Comparison between Three Types of RFID tag 9

Figure 2.3 Example of bit-tracking technology in Manchester

Coding

10

Figure 2.4 ILCM tag estimation equation parameters and

definition

14

Figure 2.5 Simulation results of FSA, DFSA, EDFSA and

FDFSA

16

Figure 2.6 Comparison of DS-MAP, SUBF-DFSA, MAP,

FEIA, ILCM and Q-algorithm

17

Figure 2.7 Efficiency of proposed algorithm and DFSA

algorithms

18

Figure 3.1 Flowchart of project implementation 20

Figure 3.2 Tag distribution process 21

Figure 3.3 Slot reservation code generation for 5 tags 22

Figure 3.4 Collision bits detection process by reader using

Manchester Coding

24

Figure 3.5 Collision bits detection process after slot

reservation code regeneration

25

Figure 3.6 Tag redistribution process 26


Faculty of Information and Communication Technology (Perak Campus), UTAR. ix

Figure 3.7 Timing details for successful, empty and

collision slots in Gen2

28

Figure 3.8 Gen2 standard parameters and description 29

Figure 4.1 Image logo of MATLAB 32

Figure 4.2 Flowchart of FSA and DFSA simulations 33

Figure 4.3 Implementation of ILCM tag estimation 35

Figure 4.4 Average time slot used in FSA, DFSA and

Proposed Approach

36

Figure 4.5 System efficiency of FSA with different frame

sizes

38

Figure 4.6 System efficiency of DFSA with different tag

estimation algorithms and Manchester Coding

39

Figure 4.7 Scenario 1 tag identification rate of FSA, DFSA

and Proposed Approach

41



42



42


Faculty of Information and Communication Technology (Perak Campus), UTAR. x

LIST OF ABBREVIATIONS

DFSA Dynamic frame slotted ALOHA

DS-MAP Dynamic Sub-frame based Maximum a

posteriori probability

EDFSA Enhanced dynamic framed slotted ALOHA

EPC Electronic Product Code

FDFSA Fitted Dynamic Framed Slotted ALOHA

FSA Framed slotted ALOHA

Gen2 Generation 2

HF High Frequency

IC Integrated Circuit

ILCM Improved Linearized Combinatorial Model

LOS Line Of Sight

LF Low Frequency

RF Radio Frequency

RFID Radio-Frequency Identification

RN16 16-bit Temporary ID

SUBF-DFSA Sub-frame DFSA

UHF Ultra-High Frequency

Chapter 1: Introduction


Faculty of Information and Communication Technology (Perak Campus), UTAR. 1

CHAPTER 1: INTRODUCTION

1.1 Problem Statement and Motivation

Due to its contactless nature and faster read rate, RFID system has become

popular and successfully attracted worldwide attention in supply market. However,

RFID system is still consists of some limitations to be solved especially tag collision

that occurred when more than one tag is transmitting data simultaneously to a reader.

Consequently, the reader cannot rightly receive the tag information and lower the tag

identification accuracy.

Among of different existing ALOHA-based anti-collision algorithms, DFSA is

always adopted to resolve RFID tag collision problem. This is due to its advantage

which is able to provide frame size that corresponding to the number of tags. But,

DFSA is heavily relying on the result of tag estimation to perform frame size

adjustment. Hence, accurate tag estimation is crucial in DFSA. This is because when a

smaller frame size is offered, it would result in increasing the number of collision

slots. Meanwhile, the number of empty slot would become higher when a bigger

frame size is offered. Therefore, selecting an optimal frame size in DFSA is always

not an easy task.

Besides, Generation 2(Gen2) protocol which involved timing during tag-

reader communication is also adopted in this project. This is because we want to study

the slot duration in Gen2 during its identification process and further shorten its

timing during identification by introducing a new timing concept. In this project, we

were going to utilise Manchester Coding which is a bit-tracking technology to

mitigate the tag collision problem. Manchester Coding is very useful in collision

detection during tag identification process as it could inform the reader the collision

occurrence whenever it detects the position of collision bits from the information

carried by RFID tags. Thus, it could help to improve system efficiency of RFID

system as the reader could notice the collision and resolve it in a shorter period of

time. As a result, it could help to reduce the time needed in tag identification process.

In a nutshell, the motivation of this project is to mitigate the problem of tag, maximise

the rate of tag identification and shorten the slot duration of RFID system based on

DFSA.




1.2 Project Background

RFID is an identification system that uses electromagnetic fields to offer

wireless communication between reader and tags. Due to its contactless nature, RFID

has largely adopted in our daily routine and industries for fast object identification

and tracking. Even with its fast identification and wireless nature, it also comes with

several constraints. As discussed in previous section, RFID system always suffers

from the problem of collision.

There are two different categories of RFID collision which are reader collision

and tag collision. Reader collision occurred when one tag is read by multiple readers

while tag collision happened when there is more than one tag sending signals to a

reader at the same time. As a result, it prolongs the tag identification time as the

reader could not recognise tags instantly. This is because the reader would need to

solve the collision via anti-collision algorithm or retransmit the command. Figure 1.1

shows the two main types of collision in RFID system.

(a) Tag collision (b) Reader collision

Figure 1.1: Diagram of RFID system collision types (slideplayer.com, n.d.)

The main focus of this project is resolving RFID tag collision problem. There

are two types of anti-collision algorithms had been proposed to encounter this

problem which are ALOHA-based and Binary Tree. In this project, we are focusing

on ALOHA-based anti-collision algorithms. The related existing works of this

algorithm are Pure ALOHA, slotted ALOHA, FSA and DFSA algorithm.

In earlier time, FSA was usually adopted to resolve tag collision. This is

because the previous Pure ALOHA and slotted ALOHA are not able to resolve the tag

effectively and efficiently when there is huge number of tags involved in reader-tag

communication. Therefore, FSA that enables the tags to send their data in random




slots within a frame and the collided tags will be identified in future frames instead of

competing with each other for the available time slots. This could reduce more

collisions as compared to both of these algorithms. Figure 1.2 illustrates an example

of FSA tag identification process.

Figure 1.2: Example of FSA tag identification process

However, FSA is always using the same frame size throughout the whole

identification process. This will lead to high collision rate and long identification time

if an improper frame size is adopted. Hence, DFSA which able to provide dynamic

frame size according to collision status of current frame is more preferable to use in

resolving RFID tag collision. However, there is one issue arise while implementing

DFSA which is selecting an optimal frame size. If an inappropriate frame size is

selected, it will affect the rate of tag identification and system efficiency of RFID

system. As a result, DFSA requires an accurate tag estimation algorithm in order to

provide optimal frame size. Due to this reason, many tag estimation algorithms such

as Lowbound, Schoute and Vogt’s algorithm and other improved estimation

algorithms had been introduced in order to tackle this problem.




1.3 Objectives

The overarching aim of this project is to mitigate tag collision problem when

there is large volume of tags involved in the tag identification process. We are going

to apply DFSA with Manchester Coding which is a bit-tracking technology in our

proposed method in order to introduce a new timing concept which would provide

the position of collision bits during RFID reader-tag communication process. By

providing this kind of information, it would help to enhance the tag identification

process of RFID system by accelerating the process of resolving tag collision.

Unlike conventional ALOHA-based anti-collision algorithms, our proposed

approach would adjust the frame size based on the Manchester Coding collision

detection results in each time slot in a read cycle and it would resolve the tag

collision slot by slot. This could help to improve the system efficiency of RFID

system as it could reduce the number of collided tag involved in each future read

cycle and thus the time slot used in the tag identification process would be lower.

The objectives of this project had been identified and listed as below:

i. Study and simulate FSA and DFSA anti-collision algorithms

ii. Study and apply ideal tag estimation, Schoute and ILCM tag estimation

algorithm in DFSA to select optimal frame size

iii. Understand timing and slot duration of Gen2 standard in RFID system

iv. Propose a prototype scheme which:

a. utilise Manchester Coding to resolve tag collisions during tag

identification process

b. shorten the timing in Gen2 standard




1.4 Proposed Approach

Figure 1.3 shows the system flow diagram of the proposed approach of this project.

Figure 1.3: Proposed Approach Flow Diagram

For the first step of the proposed approach, it would be used to define the

initial frame size that used for tag identification process. This step is the crucial part

of the proposed approach as it would provide the number of time slot that available

for the tags to reserve for identification process.

StartStart

Frame size initialisationFrame size initialisation

Tag distributionTag distribution

Slot reservation code

generation


generation

Identify success, collision and

empty slots


empty slots

Number of collided tag = 0?Number of collided tag = 0?

Gen2 timing implementationGen2 timing implementation

EndEnd

Frame = New frame sizeFrame = New frame size

Yes

No

Tag estimationTag estimation

Duplicate slot reservation

code detected?


code detected?

Yes

No




Step 2 is mainly used to perform tag distribution which allows each tag to

reserve a time slot through the random number generated. In the proposed approach,

each tag is required to have a slot reservation code in order to inform the reader

regarding their slot reservation information.

Step 3 is designed to generate an 8-bit binary slot reservation code after all the

tags have selected the time slot to be reserved. The actual time slot allocation and

collision detection of reader in our proposed approach would heavily rely on this 8-bit

binary slot reservation code.

Step 4 will be used to identify success, collision and empty time slots. In this

step, we would apply Manchester Coding to identify success and collision slots and

provide reader the actual time slot allocation decision for each tag. If any collision

bits are found in a particular slot, the tags which reserved this slot would need to

return to step 2 in order to reserve a new time slot. In contrast, if no collision bit is

found, the reader would grant the tags its reserved time slot. Hence, this step is

important for the reader to perform the actual time slot allocation and determine the

system efficiency of the proposed approach. Besides, step 4 is also designed to

prevent another type of collision which caused by the same reservation code used by

the tags to reserve a time slot.

For step 5, it would be used to generate a new frame size in future read cycle

for the tags which unable to obtain reserved time slot in previous step. All the collided

tags would go through the slot reservation process again and thus a new frame size is

created in order to provide time slots that are available for reservation.

The last step would be used to study and determine the tag identification rate

and timing used by the proposed approach throughout the tag identification process by

applying Gen2 standard timing parameters.




1.5 Report Organisation

This report consists of five chapters and is organised as following:


This chapter consists of problem statement and motivation, project background

information, objectives and the basic idea of the proposed approach.

Chapter 2: Literature Review

This chapter consists of literature reviews of technology used in this project and

discussion of previous related works.

Chapter 3: System Design

This chapter would discuss about the system design of the proposed approach and it

also consists of flow diagram and detailed steps that describing the implementation of

the proposed approach.

Chapter 4: Implementation and Analysis

This chapter would consist of the design specification of this project such as hardware

and software used and experiment design of the proposed approach. Besides, it also

includes the analysis and discussion for the simulation results and performance of the

proposed approach with previous related works.

Chapter 5: Conclusion

This chapter would consist of the project review, discussion, the contributions of the

proposed approach and future works to be done.




CHAPTER 2: LITERATURE REVIEW

2.1 Review of technologies

2.1.1 RFID system

RFID is an identification system that utilising electromagnetic fields to

automatically identify tags attached to objects. According to McDowell (2009), there

are four basic components in RFID systems which are tags, reader, antenna and

computer system for data collection and processing. A RFID reader consists of a

transceiver and an antenna. The transceiver will generate radio signal and transmit it

through the antenna. This signal is used as a form of energy to activate the tags. Later,

the RFID tag will receive the signal and the transponder will convert the radio

frequency into usable power to send message back to reader. Then, the reader will

receive the radio waves sent by transponder and interpret the radio frequencies as

meaningful data. Lastly, the reader will send the information to the host computer for

interpreting and processing. Figure 2.1 illustrates how a RIFD system works.

Figure 2.1: Diagram of how a RFID system works (lakshmi and profile, 2012)

RFID system consists of three different kinds of tags. The first one, passive

tags which do no internal power source and usually relying on the radio signal that

generated by reader to power up. Therefore, passive tags always have shorter read

range when compared to active tags and required high signal strength for

communication. Unlike passive tags, active tags have their own power source that

enables them broadcast signal and have longer read ranges. They could be also act as

“beacons” to initiate a communication with reader or other tags. In this project, we are

only focusing on the passive tag collision in RFID system.




For semi passive tag, it is a kind of passive tag that contains the feature of

active tag. Like active tag, it has internal power source that used to power itself up.

Besides, the internal power source of semi passive tag will be activated and powers

integrated circuit (IC) while a reader radio frequency (RF) signal is received. Figure

2.2 shows comparison between three types of RFID tag.

Figure 2.2: Comparison between Three Types of RFID tag (content, 2008)

2.1.2 Generation 2 (Gen2) technology

Gen2 is the first air-interface protocol that introduced by EPCglobal for ultra-

high frequency (UHF) band in 2004. It is operating in UHF frequency range of

860MHz to 960MHz. Most of the recent RFID systems are adopting Gen2 technology

as it allow the tags that do not have own power source to have the ability to reach the

reader up to the distance of 10 meters. Similar to RFID passive tags, Gen2 tags are

activated by utilising the reader generated radio waves that transmit through the

antenna. As Gen2 tags do not powered by batteries, they required an energy-efficient

technique to perform tag-reader communication. Hence, Medium Access Control

(MAC) layer that allows energy-efficient multiple tags communication is applied in

Gen2 technology.

Besides, Gen2 technology is always worked with DFSA based on Q-algorithm

for the purpose of improving the system efficiency of RFID system. The frame size in

Gen2 is usually initialised in the range of 0 to 2Q-1 and Q is an integer which is

ranging from 0 to 15 and it is broadcast by reader through Query command. The

reader is able to change the Q parameter based on the collision condition in current

frame by issuing an adjust command.




2.1.3 Manchester Coding

In this project, Manchester Coding would be utilised to resolve the problem of

tag collision. Manchester Coding is a bit-tracking technology that allows RFID reader

to recognise the location of collision bits within a collision slot. Moreover, it uses

voltage level transition to represent the value of a bit. If there is a positive transition,

the value of bit will be equal to “0” and “1” is for negative transition. Manchester

Coding allows individual bit tracing when there are two or more tags are sending

different bits and caused tag collision. Figure 2.3 illustrates an example of bit-tracking

in Manchester Coding.

Figure 2.3: Example of bit-tracking technology in Manchester Coding (Landaluce,

Perallos, and Angulo, 2014)

As showed in Figure 2.3, there are two tags transmitting their tag identifier

which are 0100110 and 0101111 respectively. However, the reader does not manage

to receive the correct tag information as the signals sent by these two tags are being

interfered. As a result, the reader is receiving 010X11X and X is denoted as the

collision bits. There is no voltage level transition when there is a collision occurred. In

the given example, the collision bits are located the 4th

and 7th

time slots. By

providing this information, Manchester Coding could help to accelerate the tag

identification process and lower the rate of tag collision.




2.2 Review of -based anti-collision algorithms

Throughout the years, many RFID anti-collision algorithms have been proposed by

researchers as it is crucial to enhance the performance of RFID system. We are going

to discuss different ALOHA-based anti-collision algorithms in the section below.

2.2.1 Pure ALOHA

Pure ALOHA is the first algorithm that had been introduced to encounter the problem

of tag collision. In this algorithm, a tag will simply transmit data to a reader whenever

it has data to be sent. However, the retransmission time needed is very long when

there is tag collision between two or more tags. The problem become worse if there is

large volume of tags sending signal to a reader at the same point of time. Thus, the

maximum throughput of Pure ALOHA is only 18.4%. (Raja and Perumal, no date)

2.2.2 Slotted ALOHA

In Slotted ALOHA, it consists of time slots that divided from the data transmission

time. Each available tag is required to select one slot to their transmit data. (Cheng

and Jin, 2007) Thus, the collision interval in Slotted ALOHA is halved as compared

to Pure ALOHA. However, the efficiency of this approach is degraded if there is huge

number of tags involved in data transmission. The maximum throughput of Slotted

ALOHA is 36%. (Raja and Perumal, no date)

2.2.3 Frame Slotted ALOHA (FSA)

In order to encounter the limitation of slotted ALOHA, FSA that consists of frame

that make up by a group of time slots has been introduced. In FSA, the tags are

randomly distributed to the time slots. The time slot is considered to be a successful

slot when only one tag occupied that slot. A collision slot would be created when

there is two or more tags select one time slot at the same time. These tags would then

transmit their data in the next read cycle. A time slot becomes an empty slot when

there is no tag selects it. This process will keep on going till all the tags are

successfully identified. However, system efficiency of FSA would influence badly by

the frame size as the frame size in FSA is remaining unchanged throughout the

identification process because it has no way to know the number of unread tags.




2.2.4 Dynamic Slotted ALOHA (DFSA)

The major limitation of FSA is its static frame size during the tag identification

process. Hence, the frame size provided might be smaller or bigger than the number

of tags. Thus, DFSA was introduced to encounter this drawback by enabling the

adjustment of the frame size during identification process. Hence, it has lower number

of used time slots as compared to FSA. The time duration used by the reader to

identify the tags is also shorter. However, system efficiency of DFSA is always

affected by the frame size offered. This makes tag estimation process become very

important in DFSA as the fame size offered is depend on the results of tag estimation

process. However, there is no standardised tag estimation algorithm available in

current stage.

2.2.5 Summary of ALOHA-based anti-collision algorithms

The table below shows the summary of different ALOHA-based anti-collision

algorithms.

ALOHA-based Anti-

collision algorithm

Strength Weakness

Pure ALOHA Easy to implement Collision become higher as

the number of tags become

bigger

Slotted ALOHA Reduce collision

interval to half

Enhance throughput

Efficiency is degraded

when the number of tags

become bigger

FSA Increase throughput and

reduce the rate of collision

Number of tags cannot be

recognised

DFSA Number of tags could be

known through estimation

Tag estimation algorithm

is not standardised

Table 2.1: Summary of ALOHA-based anti-collision algorithms




2.3 Review of existing tag estimation algorithms

After reviewing the previous related works, the anti-collision algorithm that we are

going to adopt in this project is DFSA. However, selecting an optimal frames size in

DFSA is a very challenging task. In this section, we are going to discuss different

existing tag estimation methods that assist DFSA in selecting optimal frame size.

2.3.1 Lowbound algorithm

In Lowbound algorithm, it will predict the number of unknown tags bay assuming that

there are two or more collision tags. Therefore, it predicts the number of tags using

the formula S + 2C where C represents the number of collision slot and S represents

successful slots within one frame. However, the tag estimation error increases when

there are more than two collision tags.

2.3.2 Schoute algorithm

Schoute algorithm was using Poisson distribution to obtain the expected number of

collision slots. Psucc and Pcoll is the probability of success and collision occurred in a

time slot respectively. The formula used by Schoute to predict the number of

unknown tags is S + 2.39C. Therefore, tag estimation of Schoute is more efficient

than Lowbound. However, Schoute algorithm is having the same drawback as

Lowbound algorithm as they are doing estimation without considering the actual

collision condition of current frame. Thus, it would have large estimation error when

there is large number of collision tags.

2.3.3 Improved Linearized Combinatorial Model (ILCM)

ILCM is a tag estimation algorithm with low computational cost and was

introduced by Solic et al.(2013). ILCM is a scheme that performs frame break when

the next frame has higher expected number of successful slots than current frame. The

tag estimation of ILCM is done through 𝑝(𝐸, 𝑆, 𝐶|𝑛) = 𝐿!

𝐸!𝑆!𝐶!

𝑁𝑆(𝑛,𝑆)𝑁𝐶(𝑛,𝑆,𝐶)

𝐿𝑛 where

frame size, L is equal to E+S+C. Figure 2.4 shows the parameters that involved in

ILCM tag estimation equation and their definition.




Figure 2.4: ILCM tag estimation equation parameters and definition (Solic et al.,

2013)

However, this relation is computationally heavy as it needs both hardware and

software to perform the calculation. Thus, it was later simplified into �̂� = 𝑘𝑆 + 𝑙 to

reduce the computational cost. In this project, we are going to adopt ILCM as one of

tag estimation algorithm used during tag estimation process.

2.3.5 Summary of existing tag estimation algorithms

Table 2.2 shows the summary of different existing tag estimation algorithms.

Tag estimation

algorithm

Strength Weakness

Lowbound Easy to implement Tag estimation error increases

when collision tag is more than

two

Schoute More accurate tag estimation Tag estimation error increases

when collision tag is more than

two

ILCM Low computational cost Complex calculation

Table 2.2: Summary of existing tag estimation algorithms




2.4 Review of existing improved ALOHA-based anti-collision algorithms

As existing ALOHA-based anti-collision algorithms could not resolve tag collision

problem effectively, numerous related improved works has been proposed throughout

the years. Out of the many improved related works, three relevant ones were selected

to be reviewed in detail as follows.

2.4.1 Fitted Dynamic Framed Slotted ALOHA Anti-Collision Algorithm in RFID

Systems

In Shakiba, Zavvari, and Sundararajan paper (2011), their proposed method,

Fitted Dynamic Framed Slotted ALOHA (FDFSA) was to shorten the tag

identification time by using minimum slots number. The proposed algorithm consists

of four parts.

The first part which is the main part of the proposed algorithm is to define an

initial frame size. This is to initiate a read cycle. Next, all tags would be assigned to

different time slots based on the random number generated by distribution function.

After that, the tags would send their IDs to reader. In next step, the slots would be

read one by one after calling the read function. A tag is successfully identified when it

is a successful slot and the tag will also assign a number of -1. This read function

would count the number of successful slots (Cl) and collision slots (CK).

Lastly, curve fitting estimation function is called to predict the number of tags

according to Cl and CK. If there is a collision occurred in current read cycle, these tags

would be identified in future read cycle with a new frame size which is created based

on tag estimation results. This process would be stopped when all the tags are

successfully identified. Shakiba, Zavvari, and Sundararajan had compared and

evaluated the performance of FDFSA with curve fitting estimation with other existing

algorithms such as FSA, DFSA and enhanced dynamic framed slotted ALOHA

(EDFSA) using the total number of slots used during tag identification with the initial

frame size of 64. Figure 2.5 shows the simulation results of FSA, DFSA, EDFSA and

FDFSA.




Figure 2.5: Simulation results of FSA, DFSA, EDFSA and FDFSA

(Shakiba, Zavvari, and Sundararajan, 2011)

From their simulation results, they had concluded that FDFSA used the least

number of required slots in tag identification process as compared to other algorithms.

But, their proposed method did not consider the number of idle slots and thus

efficiency of the proposed method would be reduced when there is high number of

idle slots.

2.4.2 An Efficient and Easy-to-implement Tag Identification Algorithm for UHF

RFID Systems

In this paper, a sub-frame based DFSA algorithm, Dynamic Sub-frame based

Maximum a posteriori probability (DS-MAP) was proposed with the purpose of

improving the tag identification efficiency of RFID systems. In this proposed method,

it will not perform tag estimation calculation in the reader itself but utilise tables to

pre-store the estimation results. By looking up the tables, it could reduce MAP

computation overhead which caused by nested loop. However, it might require more

memory to store the tables when the number of trials when n tags compete for slots.

Thus, sub-frame structure is used in order to save memory and limit the table size.

In DS-MAP, it would divide a frame into sub-frames and assume the

estimated tag numbers are equal in each sub-frame under the condition that all the

tags ware evenly distributed. By referring to the number of successful and collision

slots, it will predict the number of tags in first sub-frame by looking up the pre-stored




DS-MAP tables using the formula, 𝑁𝑒𝑠𝑡 = �̂� ∗ 𝐾 where 𝐾 = 𝐹/𝐹𝑠𝑢𝑏 and 𝐹

represents the frame length. If the estimated number of tags fits the current frame

length, the algorithm will return to original DFSA and the frame length would be

considered as optimal. As a result, the chances of new frame size fits to the number of

backlog will be higher.

If the estimated number of tags did not fit the current frame length, it would

adjust the frame length and calculate the estimated number of backlog using the

formula 𝑁𝑏𝑎𝑐𝑘 = 𝑁𝑒𝑠𝑡 − 𝑁𝑆 where 𝑁𝑆 denotes the number of successful slot in sub-

frames. After that, the new frame length will be determined by the reader based on

𝑁𝑏𝑎𝑐𝑘 and then it will issue a QueryAdj command to update the frame length. Chen,

Su, and Yi (2017) had compared their proposed algorithms with SUBF-DFSA, MAP,

FEIA, ILCM and Q-algorithm. Figure 2.6 shows the comparison of these algorithms.

Figure 2.6: Comparison of DS-MAP, SUBF-DFSA, MAP, FEIA, ILCM and Q-algorithm

(Chen, Su, and Yi, 2017)

From figure 2.7, we could note that the proposed method is the second least

computation complexity as it only needs to look up tables in order to count the

number of success, idle and collision slots. DS-MAP also required lesser memory to

store the tables. However, the significant drawback of this proposed method was it

trade-off the computation complexity that would largely affect the energy

consumption and RFID tag identification time with memory size.




2.4.3 A Dynamic Framed Slotted ALOHA Anti-collision Algorithm Based on

Tag-Grouping for RFID Systems

In this paper, the authors had introduced a tag-grouping method that RFID

reader identifies the tags group by group and it was divided into two main parts which

were randomisation grouping process and collision slots identification process. Firstly,

an initial frame length, L = 2 is set and the reader would send this parameter with a

query command. Then, all the tags would randomly select a slot from 0 to L-1 after

receiving the query command and then transmit reader a 16-bit random number in the

corresponding slot. Next, the reader would identify success slot, idle slot and collision

slot when reader received all the tag responses.

The next operation that would be performed by the reader is based on the

response slots. The reading process would be terminated if all the response slots are

idle slot. If the response slots are single slots then the reader will read the tags in this

kind of slot and these tags would exclude from the reading process. If the responded

slots are collision slots, firstly, the reader adjust the frame length become L * 2 and

send this parameter with a query command. Then, all the unread tags will randomly

select a slot in range of 0 to 3. The reader could read all the tags if there is no collision

occurred and stop the reading process. If collision occurs, reader will transmit a new

query command with a parameter which double the number of collision slots in order

to read other tags. Figure 2.7 shows the efficiency of proposed algorithm and DFSA

algorithms.

Figure 2.7: Efficiency of proposed algorithm and DFSA algorithms (Qing et al., 2012)




Qing et al. had designed 50 experiments by setting the initial frame length to

16 with 0 to 1000 tags in order to compare their proposed algorithm with other DFSA

algorithms. From Figure 2.8, we could observe that the efficiency of their proposed

algorithm was able to achieve up to 35% which is the highest among other DFSA

algorithms. However, in order to reach highest efficiency of this proposed algorithm,

a probability of 0.9 containing one tag in at least one slot in a frame needs to be

achieved which is very biggest challenge faced by their proposed method.

2.4.4 Summary of existing improved ALOHA-based anti-collision algorithms

The following table shows the summary of different existing improved works that has

been studied in the literature review.

Existing improved

anti-collision

algorithm

Strength Weakness

Fitted Dynamic

Framed Slotted

ALOHA Anti-

Collision Algorithm

(FDFSA)

Shorten tag identification

time

Consume lesser time slots

Number of idle slots is not

considered and thus the

efficiency of the algorithm

would be reduced when there

is large number of idle slots

Dynamic Sub-frame

based Maximum a

posteriori probability

(DS-MAP)

Able to adjust the frame

length to fit the tag number in

shorter time

Complex computation

Higher energy consumption

Slower tag identification

speed

DFSA based on Tag-

grouping

Able to disperse tags

quickly and evenly

Enhance the efficiency of

RFID systems

Probability of 0.9 containing

one tag in at least one slot in a

frame needs to be achieved in

order to reach highest

efficiency

Table 2.3: Summary of existing improved ALOHA-based anti-collision algorithms




CHAPTER 3: SYSTEM DESIGN

3.1 System flow

In this project, we were going to apply DFSA that could adjust the frame size

dynamically based on the collision situation in our proposed approach together with a

bit-tracking technology, Manchester Coding to mitigate the tag collision problem

occurred in RFID system. Figure 3.1 shows the implementation flowchart of the

proposed approach of this project.

Figure 3.1: Flowchart of project implementation

StartStart

Frame size initialisationFrame size initialisation

Tag distributionTag distribution


generation


generation


empty slots


empty slots

Number of collided tag = 0?Number of collided tag = 0?

Gen2 timing implementationGen2 timing implementation

EndEnd

Frame = New frame sizeFrame = New frame size

Yes

No

Tag estimationTag estimation


code detected?


code detected?

Yes

No




Step 1: Frame size initialisation

The first step of RFID tag identification process is to initialise the frame size,

L used to identify the tags. In the proposed approach implementation, we always

assumed that the initial frame size for DFSA is always equal to the number of tags. As

we want to provide frame size that is optimal for the tag identification process. For

instance, if there are five tags to be identified by the reader then L would be set equal

to 5.

Step 2: Tag distribution

In this step, each tag would generate random numbers within the range of

frame size via randi() function. These generated random numbers are representing the

time slot that would be reserved by the tags. For example, when L = 5 and then the

possible random numbers that would be generated by the tags is ranging from 1 to 5.

The tag distribution process is illustrated in Figure 3.2.

Figure 3.2: Tag distribution process

Tag Time slot to be reserved

1 2

2 2

3 3

4 5

5 5

Slot 1 Slot 2 Slot 3 Slot 4 Slot 5

Tag 1

Tag 2

Tag 3

Tag 4

Tag 5

Frame




Step 3: Slot reservation code generation

After selecting the time slot to be reserved, each tag would generate a random

number in the range of 1 to 255 using randi() function and convert to an 8-bit binary

bit string using de2bi() function. Then, these 8-bit binary codes would be sent to the

reader by the tags and this completes the slot reservation process. The reader would

detect the collision bits from these slot reservation codes via Manchester Coding in

the next step. This is to detect whether any collision is happened before the reader

allocates the reserved time slot to the tags. The slot reservation code generation for

five tags is showed in Figure 3.3.

Figure 3.3: Slot reservation code generation for 5 tags

Step 4: Identify success, collision and empty slots

After the tags reserving the time slots using its slot reservation code generated

in previous step, the reader would categorise the time slots into success, collision and

empty slot. In this step, we are going to apply Manchester Coding to identify success

and collision slots.

A time slot would be treated as a success slot if there is one tag reserve that

time slot and its slot reservation code could correctly receive by the reader. Then, the

reader would allocate that particular time slots for these tags. The tags in success slot

would be successfully identified by the reader and store in success list in the

following step. Meanwhile, when there is no tag reserve a particular time slot then

that time slot would become an empty slot. All the empty slots would not go through

the collision bits detection process and the reader would not receive any slot

reservation code from these slots.

On the other hand, when the reader could not receive the slot reservation codes

correctly and that particular time slot would become a collision slot. This

Tag Reserved

Time Slot

Generated Random

Number

8-bit slot reservation code

1 2 5 0 0 0 0 0 1 0 1

2 2 10 0 0 0 0 1 0 1 0

3 3 6 0 0 0 0 0 1 1 0

4 5 1 0 0 0 0 0 0 0 1

5 5 1 0 0 0 0 0 0 0 1




phenomenon occurs whenever there is a time slot reserved by more than one tag. As a

consequence, these tags could not be granted the reserved time slot from the reader

and stored in collision list and need to reserve new time slots in future read cycle with

a new frame size.

Besides, there is another type of collision that occurred when there are more

than one tag is using the same slot reservation code to reserve a particular time slot. In

this case, the reader might mistreat that slot as a success slot as there is no collision

bits found during Manchester Coding collision bits detection process. In order to

resolve this kind collision, a new slot reservation code would be generated for a tag

when there is more than one tag using the same slot reservation code to reserve a time

slot. After regenerating new slot reservation codes, these tags would go through

Manchester Coding collision bits detection process once again in order to mark it as a

collision slot.

Step 4.1: Identify empty slots

In step 4, firstly, we would identify which time slot is not reserved by any tags

to be identified by the reader. Then, we would mark this type of slot as empty slot and

these slots are not going through Manchester Coding collision bits detection process.

Thus, the reader would not receive any slot reservation code from these slots.

Meanwhile, those time slots which were reserved by the tags in previous step would

go through the collision bits detection process in order to find out which time slot is

suffering collision. From Figure 3.2, we could notice that Slot 1 and 4 are empty slots

and hence these two slots will skip all the following steps.

Step 4.2: Detect collision bits using Manchester Coding

After identifying empty slots, we would perform Manchester Coding collision

bits detection for the time slots which were reserved by tags in previous step. In order

to identify collision bits using Manchester coding, we would obtain the number of bit

‘1’ of tags’ slot reservation codes in each slot by using sum() function. Before that,

we would identify the number of tags in each time slot. Thus, if the checking result is

either less than the number of tags that reserved that particular time slot or not equal

to 0, we could know that there is a collision bit detected. Figure 3.4 illustrates the

reader collision bits detection process.




Reader Manchester Coding collision bits detection process:

(Empty) (Collision) (Success) (Empty) (Success)

Figure 3.4: Collision bits detection process by reader using Manchester Coding

Step 4.3: Check duplicate slot reservation code

Instead of using Manchester Coding to detect collision, we would also acquire

collision slots information through the number of tags in each time slot. This is done

because we need this information to resolve the collision which caused by the same

slot reservation code used by two or more tags to reserve a time slot. Besides, we

would also check the tags’ slot reservation code in each time slot. Thus, when there is

more than one similar slot reservation code is used to reserve a time slot, then all the

tags which reserved that particular time slot would be required to regenerate a random

number using randperm() and convert to 8-bit slot reservation code using de2bi().

From Figure 3.4, we could notice that Slot 5 was mistreating as a success slot

as there is no collision bits detected. This is because both tag 4 and 5 were reserving

Slot 5 by using the same slot reservation code. However, by referring to Figure 3.3,

we could know that Slot 5 is collision slot as there is more than one tag were

reserving it. Hence, tag 4 and 5 would need to regenerate a new slot reservation code

before proceeding to the next step. Besides, the reader would perform collision bits

detection process once again in order to mark Slot 5 as collision slot. Figure 3.5

shows the collision bits detection process after slot reservation code regeneration.


Tag 1 00000101

Tag 2 00001010

Tag 3 00000110

Tag 4 00000001

Tag 5 00000001

Check Result - 00001111 00000110 - 00000002

Reader - 0000???? 00000110 - 00000001

Frame 1

Read Cycle 1




Slot reservation code regeneration for tags in Slot 5:

Reader Manchester Coding collision bits detection process after slot reservation

code regeneration:

(Empty) (Collision) (Success) (Empty) (Collision)

Figure 3.5: Collision bits detection process after slot reservation code regeneration

Step 5: Tag estimation

After identifying success, collision and empty slots using Manchester Coding,

the tags which not managed to obtain their reserved time slot would go to next read

cycle with a new frame. In DFSA, the frame size would be adjusted before going to

the next read cycle and hence tag estimation that provide the estimated number of

unknown tags would be taken place. In our proposed approach, we are going to apply

ideal tag estimation which the frame size is always equal to the number of tags.

In this step, we would also generate a success and collision list to store the

tags which are either successfully or failed to recognise by the reader. If the number

of tags stored in collision list is not equal to zero, then we would redistribute new time

slots for these tags in future read cycle with a new frame size. If there is any collision

is detected during tag redistribution process, these tags will repeat the whole slot

reservation process until it is successfully read by the reader. In our proposed

approach, the tag redistribution process would start from the first collision slot that

Tag Reserved

Time Slot

New Generated

Random Number

New 8-bit slot reservation code

4 5 155 1 0 0 1 1 0 1 1

5 5 65 0 1 0 0 0 0 0 1


Tag 1 00000101

Tag 2 00001010

Tag 3 00000110

Tag 4 10011011

Tag 5 01000001

Check Result - 00001111 00000110 - 11011011

Reader - 0000???? 00000110 - ??0??0??




stored in collision list and perform tag estimation to create new frame. In our

proposed approach, the collision occurred in first collision slot must be resolved

before proceeding to the next collision slot. Figure 3.6 shows the tag estimation and

redistribution process of the proposed approach.

Success List:

Collision List:

Tag redistribution for 1st collision slot:

Number of tags in Slot 2 = 2, New frame size = 2

(Empty) (Collision) (Success) (Empty) (Collision) (Success) (Success)

Tag Reserved Time Slot 8-bit slot reservation code

3 3 0 0 0 0 0 1 1 0


1 2 0 0 0 0 0 1 0 1

2 2 0 0 0 0 1 0 1 0

4 5 1 0 0 1 1 0 1 1

5 5 0 1 0 0 0 0 0 1

Tag New Reserved Time Slot 8-bit slot reservation code

1 1 0 0 0 0 0 1 0 1

2 2 0 0 0 0 1 0 1 0

Slot 1 Slot 2 Slot 3 Slot 4 Slot 5 Slot 1 Slot 2

Tag 1 00000101 00000101

Tag 2 00001010 00001010

Tag 3 00000110

Tag 4 10011011

Tag 5 01000001

Check Result - 00001111 00000110 - 11011011 00000101 00001010

Reader - 0000???? 00000110 - ??0??0?? 00000101 00001010

1st collision

slot

Frame 1

Read Cycle 1

Frame 2

Read Cycle 2




Updated Success List:

Updated Collision List:

Figure 3.6: Tag redistribution process

From Figure 3.6, we could observe that the first collision slot is Slot 2 which

was reserved by tag 1 and 2. Thus, the collision occurred in Slot 2 would be resolved

first before proceeding to Slot 5. From the collision list, we could notice that there are

two tags reserved Slot 2 and thus by using ideal tag estimation we would set the value

of new frame size as 2. Both tag 1 and 2 would reserve a new time slot by generating

a random number in range of this new frame size which from 1 to 2.

After that, the reader would detect collision bits via Manchester Coding once

again. From Figure 3.6, we could notice that the new time slot which reserved by tag

1 and 2 are Slot 1 and 2 respectively and thus there is no collision bits detected and

the reader could receive the slot reservation codes correctly. Hence, the reader would

assign these two tags to their reserved time slots and insert into success list and then

remove from the collision list. However, the whole tag redistribution process would

be repeated again if there is any collision detected. This process would repeat for all

the following collision slots until the collision list becomes empty.


3 3 0 0 0 0 0 1 1 0

1 1 0 0 0 0 0 1 0 1

2 2 0 0 0 0 1 0 1 0


4 5 1 0 0 1 1 0 1 1

5 5 0 1 0 0 0 0 0 1




Step 6: Gen2 timing implementation

The last step is to implement Gen2 timing in the proposed approach in order to

study the slots duration used during tag identification process. In Gen2 protocol, the

reader will send the information about a frame using 22-bit Query and send 4-bit

Query Rep to tags at every beginning of the slot. Later, random number would be

generated by the tag and the number of QueryRep is counted. When generated

random number is same to the counted number, the tag would respond to the reader

query. As it allows the tag to send 16-bit temporary ID (RN16) during its slot time

and this can help to decrease the collision and empty slots time indirectly.

In Gen2 standard, a successful slot will be created when the reader could

successfully decode tags RN16 and acknowledge it by replying an ACK command

that comes with tag identifier, Electronic Product Code (EPC). If the tags are not

successfully identified, the reader will reply a negative acknowledge (NAK)

command and these tags would be read in following read cycles. The tag

identification process of Gen2 would not be terminated whenever there are collision

slots. Figure 3.7 illustrates the timing details for successful, empty and collision slots

in Gen2.

Figure 3.7: Timing details for successful, empty and collision slots in Gen2 (Solic et al., 2013)




In Gen2 standard there is a list of parameters involved in the tag identification

process. Figure 3.8 shows the Gen2 standard parameters and its description.

Figure 3.8: Gen2 standard parameters and description (Nov-2013, version 2.0 EPC™ radio-

frequency identity protocols generation-2 UHF RFID specification for RFID air interface

protocol for communications at 860 MHz – 960 MHz version 2.0.0 ratified, 2013)




The Gen2 Timing implementation steps are described as following:

Step 6.1 Select Tari, DR and BLF

The first step to implement Gen2 timing, we have to select Tari, DR and BLF value to

be used in the experiment from the given value range. The following table shows the

given value ranges of Tari, DR and BLF provided in Gen2 standard.

Parameter Value Range

Tari 6.25𝜇𝑠 𝑡𝑜 25𝜇𝑠

DR 64

3𝑜𝑟 8

BLF 40𝑘𝐻𝑧 ≤ 𝐵𝐿𝐹 ≤ 640𝑘𝐻𝑧

Table 3.1: Tari, DR and BLF and their value ranges

Step 6.2 Calculate Rbl, PRT, Tpri, TRCal and RTCal

After selecting the value of Tari, DR and BLF values, we are going to calculate the

value of Rbl, PRT, Tpri, TRCal and RTCal which is based on the value of parameters

in previous step. The following table shows Rbl, PRT, Tpri, TRCal and RTCal with

their equation and given value range for calculation.

Parameter Equation/Value Range

Rbl (2𝑇𝑎𝑟𝑖 + 0.5𝑇𝑎𝑟𝑖)/2 ≤ 𝑅𝑏𝑙 ≤ 3𝑇𝑎𝑟𝑖/2

Tpri 1/𝐵𝐿𝐹

TRCal 𝐷𝑅 ∗ 𝑇𝑝𝑟𝑖

RTCal 1.5𝑇𝑎𝑟𝑖 ≤ 𝑅𝑇𝐶𝑎𝑙 ≤ 2𝑇𝑎𝑟𝑖

PRT 12.5 ∗ 10−6 + 𝑇𝑎𝑟𝑖 + 2.5𝑇𝑎𝑟𝑖 + 1.1 𝑇𝑅𝐶𝑎𝑙

Table 3.2: Equations for Rbl, PRT, Tpri, TRCal and RTCal and Value Range

Step 6.3: Calculate TQuery, TACK, TQrep, T1, T2 and T3

Next, we will calculate the duration of TQuery, TACK, TQrep, T1, T2 and T3. The

equations for calculation and their given value range is showed in Table 3.3.




Parameter Equation/Value Range

TQuery 𝑃𝑅𝑇 + 22𝑅𝑏𝑙

TACK 𝑇𝐹𝑆 + 18𝑅𝑏𝑙

TQrep 𝑇𝐹𝑆 + 4𝑅𝑏𝑙

TFS 12.5 ∗ 10−6 + 𝑇𝑎𝑟𝑖 + 2.5𝑇𝑎𝑟𝑖 ≤ 𝑇𝐹𝑆 ≤ 12.5 ∗ 10−6 + 𝑇𝑎𝑟𝑖 + 3𝑇𝑎𝑟𝑖

T1 max(RTCal,10Tpri)* (0.9)-2*10-6 ≤ T1 ≤ max(RTCal,10Tpri)* (1.1)+2*10-6

T2 3Tpri ≤ T2 ≤ 20Tpri

T3 Minimum of 0.1Tpri

Table 3.3: Equations for TQuery, TACK, TQrep, T1, T2 and T3 and Value Range

Step 6.4 Calculate TRN16 and TEPC

In this step, we are going to calculate the value of TRN16 and TEPC. Before that, the

value of 𝑀and 𝑇𝑅𝑒𝑥𝑡 has to be selected. In this project, we are using 𝑀 = 1 and

𝑇𝑟𝑒𝑥𝑡0 = 4. Table 3.4 shows the equations to calculate the value of TRN16 and TEPC.

Parameter Equation

TRN16 ((𝑇𝑟𝑒𝑥𝑡0 ∗ 𝑀)/𝐵𝐿𝐹) + ((6𝑀)/𝐵𝐿𝐹) + ((17𝑀)/𝐵𝐿𝐹)

TEPC ((𝑇𝑟𝑒𝑥𝑡0 ∗ 𝑀)/𝐵𝐿𝐹) + ((6𝑀)/𝐵𝐿𝐹) + ((𝑀 ∗ (16 + 96 + 17))/𝐵𝐿𝐹)

Table 3.4: Equations for TRN16 and TEPC

Step 6.5 Calculate TS, TC and TE

The last step of Gen2 timing implementation is to calculate the duration of empty,

collision and successful slot. Table 3.5 below shows the equations to calculate the

value of TS, TC and TE.

Parameter Equation

TS 𝑇𝑄𝑟𝑒𝑝 + 𝑇1 + 𝑇𝑅𝑁16 + 𝑇2 + 𝑇𝐴𝐶𝐾 + 𝑇1 + 𝑇𝐸𝑃𝐶 + 𝑇2

TC 𝑇𝑄𝑟𝑒𝑝 + 𝑇1 + 𝑇𝑅𝑁16 + 𝑇3

TE 𝑇𝑄𝑟𝑒𝑝 + 𝑇1 + 𝑇3

Table 3.5: Equations for TS, TC and TE




CHAPTER 4: IMPLEMENTATION AND ANALYSIS

4.1 Design Specifications

In this project, experiments with different number of tags which changing from 5 to

1000 had been carried out to analyse and evaluate the performance of FSA, DFSA

and our proposed approach in Gen2 standard. This project is using MATLAB to

perform simulation among the different algorithms and produce simulation results in

graphical form for analysis purpose. The following are the minimum system

requirements for this project sorted into hardware and software categories.

4.1.1. Hardware

A) Personal Computer (PC)

Pre-installed with MATLAB.

4.1.2. Software

The software that would be equipped in this project is MATrix LABoratory

(MATLAB). The following figure shows the image logo of MATLAB.

Figure 4.1: Image logo of MATLAB

B) MATrix LABoratory (MATLAB)

MATLAB is the easiest and most productive software that provides high-level

language for numerical computation, data analysis, and application

development. (The MathWorks, Inc., n.d.)

This project involves simulation of different ALOHA-based anti-collision

algorithm and the simulation involved many calculations and parameters

during the tag estimation and identification process. Therefore, MATLAB

which includes an extensive set of built-in math functions and 2D and 3D

plotting functions is required in the project implementation in order to provide

fast mathematical calculations and a visualized data and communication

results.




4.2 Implementation of FSA and DFSA

In this project, we were also carried out the simulation for existing ALOHA

anti-collision algorithms which are FSA and DFSA. This is because we were going to

evaluate and compare the performance among FSA, DFSA and the proposed approach

in term of system efficiency and the tag identification rate. For DFSA, three tag

estimation methods which are ideal tag estimation, Schoute algorithm and ILCM were

applied. This is because we want to study the efficiency of DFSA with these three

algorithms. The following section would describe the steps involved in both FSA and

DFSA simulations. Figure 4.2 shows the flowchart the implementation of FSA and

DFSA simulations.

Figure 4.2: Flowchart of FSA and DFSA simulations

Start

End

Frame size

initialisation

Identify success,

collision and

empty slots

Tag estimation

Success slots < Number of

tags

Frame = New frame size

Tag distribution

No

Yes

Gen2 timing

implemention




Step 1: Frame size initialisation

Similar as the proposed approach, we need to initialise the frame size, L to be

used to identify RFID tags. In our simulation, the initial frame size for DFSA is

always equivalent to the number of tags that are available whereas the initial frame

sizes used in FSA are 100, 150 and 200 respectively. Hence, there would be a total of

three simulations would be done in FSA.

Step 2: Tag distribution

After defining the frame size, the tags would randomly select the time slot by

generating random numbers within the range of frame size via randi() function. The

generated random numbers are representing the time slot which would be chosen by

the tags. For instance, if there are 3 tags to be identified by reader and then the

possible random numbers that would be generated by the tags is ranging from 1 to 3.

Step 3: Identify number of empty, success and collision slots

After distributing the tags to their respective time slots, the next process is to

identify the tags either are collided or successfully identified by the reader. The slot

would be considered as successful slot when there is one and only one tag is assigned

to it. In contrast, if there is more than 1 tag are choosing the same time slots and that

time slot would become a collision slot. When there is none of the tags are assigning

to a particular time slot, the time slot would become an empty slot. For the tags which

are suffering from tag collision would not be identified and they will go to the next

reading cycle and then a new frame would be used.

Step 4: Tag estimation

Whenever there is a tag collision detected in previous step, the unread tags

would go to next read cycle. Thus, tag estimation would be taken place in order to

adjust the frame size. FSA would not go through this step as the frame size used will

remain unchanged throughout the identification process. For DFSA, we are going to

apply ideal tag estimation, Schoute algorithm and ILCM for tag estimation process. In

ideal tag estimation, we assume the frame size is always equal to the number of tags.




Step 4.1 Schoute tag estimation algorithm implementation

As mentioned in section 2.4.2, Schoute algorithm would estimate the number

of unknown tag, �̂� by using the formula �̂� = 𝑆 + 2.39 ∗ 𝐶 . Hence, the new frame

size is determined by using the formula 2.39 ∗ 𝐶 . For example, if there are two

collision slots in current frame. Then, the new frame size would be 2.39 ∗ 2 . The

result 4.78 will later round to the nearest integer which is 5.

Step 4.2 ILCM tag estimation algorithm implementation

In ILCM, the frame size, L is set to S + C + E. The estimation for unknown

tags of ILCM is done through the simplified relation, �̂� = 𝑘𝑆 + 𝑙 where

𝑘 =𝐶

(4.334𝐿−16.28)+(𝐿

−2.282−0.273𝐿)𝐶+0.2407 ln(𝐿+42.56)

𝑙 = (1.2592 + 1.513𝐿)𝑡𝑎𝑛 (1.236𝐿−0.9907𝐶). (Solic et al., 2013)

Besides, there are two scenarios that ILCM estimation would be bounded in.

The first scenario is when value of k is less than 0 is given to smaller L and then the

estimation should return k=0. Another scenario is when there involved an estimation

error where C=0. In such scenario, �̂� would set to S. Figure 4.3 show the

implementation of ILCM tag estimation.

Figure 4.3: Implementation of ILCM tag estimation (Solic et al., 2013)

Step 5: Gen2 timing implementation

The last step is to implement Gen2 timing in both FSA and DFSA in order to

study the slots duration during tag identification process.




4.3 Results and discussion

The simulation results discussion would be organised into two parts which are

comparison between FSA, DFSA and our proposed approach, DFSA with Manchester

Coding with and without Gen2 standard respectively.

4.3.1 Comparison between FSA, DFSA and Proposed Approach without Gen2

standard

In this section, we are going to compare the simulation results of FSA, DFSA

and proposed approach which is DFSA with Manchester Coding without Gen2

standard. This is to discuss and evaluate the performance of proposed approach and

the existing ALOHA anti-collision algorithms which are FSA and DFSA by

comparing the average time slot used during tag identification and system efficiency.

A. Average time slot used

The following figure presents the average time slot used in FSA (with frame size

of 200), DFSA and proposed approach with 5 to 1000 tags.

Figure 4.4: Average time slot used in FSA, DFSA and Proposed Approach

From Figure 4.4, it shows that the average of time slot used in FSA is

increased exponentially as the number of tags becomes bigger. When there are 1000

tags involved in the identification process, the time slot used is approximately 8800.

This happened is due to the initial frame size provided in the experiment which is 200

could not cater the needs of this huge number of tags. Furthermore, this frame size is

0 200 400 600 800 10000

500

1000

1500

2000

2500

3000

3500

Number of tags

Nu

mb

er

of tim

e s

lot

Average time slot used in DFSA

0 200 400 600 800 10000

1000

2000

3000

4000

5000

6000

7000

8000

9000

Number of tags

Nu

mb

er

of tim

e s

lot

Average time slot used in FSA

DFSA

DFSA with Manchester Coding




remained static throughout the identification process. If smaller frame size is

initialised, the rate of tag collision will become higher. As a result, the tags will go

through many rounds of read cycle and the number of time slot used will become

higher.

Meanwhile, we could learn that DFSA has reduced the number of average

used time slot to only around 2700 which is more than half as compared to FSA. This

is because DFSA could dynamically adjust the frame size as the collision situation

reported in the current frame. However, DFSA would resolve the tag collision

detected from different time slots in one read cycle and this process will require

higher number of time slot if there is large number of collided tags found in current

read cycle. For instance, if there are 100 collided tags detected in one read cycle,

DFSA would need to prepare 100 time slots in the consequent read cycle. This would

also create higher number of empty slots especially in the worst case which all the

tags select the same time slot.

Unlike the conventional DFSA, our proposed approach will resolve the tag

collision slot by slot. For example, when there are 3 collision slots and 10 collided

tags detected in one read cycle, our proposed approach would sort these collided tags

according to their reserved time slot. After that, it would resolve the collision

accordingly based on the slot number. For instance, if there are 3 collision slots which

are Slot 1, 2 and 3 found in current read cycle. Our proposed approach would first

resolve the collision in first collision slot which is Slot 1. If there are 3 collided tags

found in Slot 1, our proposed method would provide time slot available for

reservation in the next read cycle that fit to this number of collided tags. By doing this,

it could reduce the number of empty slot and slot used per read cycle. From Figure 4.4,

we could notice that the number of time slot used in our proposed approach when

there are 1000 tags is reduced to approximately 2300. Therefore, it could provide

higher system efficiency as compared to traditional DFSA.

B. System Efficiency

System efficiency is another main concern of selecting anti-collision algorithm.

In ideal case, maximum efficiency that could be theoretically achieved by FSA is

36.8%. However, the system efficiency of FSA would decrease gradually when an




0 100 200 300 400 500 600 700 800 900 10000

5

10

15

20

25

30

Number of tags

Syste

m E

ffic

iency (

%)

System Efficiency of FSA

L = 200

L = 150

L = 100

improper frame size is used in the identification process. Figure 4.5 shows the system

efficiency of FSA with frame size of 100, 150 and 200.

Figure 4.5: System efficiency of FSA with different frame sizes

From Figure 4.5, the maximum figure that FSA could achieve in the

simulations is ~27%. Especially when frame size, L = 100, the system efficiency of

RFID system is 0.3842% which almost 0% when there are 1000 tags. This is because

of the frame size used was smaller than the number of tags. Hence, the tags need to

compete with each other in order to obtain a time slot and this would cause high rate

of tag collision. In FSA, an optimal frame size is very important as it could influence

the efficiency of RFID system badly if the frame size is not properly defined.

Therefore, it is proven that FSA is no longer feasible to resolve tag collision as there

is no way to decide the initial frame size to be used and hence DFSA is introduced.

However, DFSA is also having the difficulty in selecting an optimal frame

size. It needs to have prior notice about the number of tags before selecting a frame

size. Hence, tag estimation is essential to resolve DFSA dilemma in choosing frame

size. In this project, the selected tag estimation algorithms to be simulated are ideal

tag estimation, Schoute algorithm and ILCM respectively. In our proposed method,

we adopted Manchester Coding collision bits detection results together with ideal tag

estimation to provide optimal frame size for tag identification process. Figure 4.6

shows the system efficiency of DFSA with ideal tag estimation, ILCM, Schoute

algorithm and Manchester Coding.




Figure 4.6: System efficiency of DFSA with different tag estimation algorithms and

Manchester Coding

As showed in Figure 4.6, DFSA is giving more stable system efficiency as

compared to FSA. The system efficiency of DFSA with ideal tag estimation, Schoute

algorithm and ILCM would able to provide a more stable system efficiency which is

~36% as the number of tags becomes larger. Besides, we could observe that ILCM

has the lowest system efficiency among other algorithms. This is because the

performance of ILCM is affected by the initial frame size. Thus, ILCM could not

adjust the frame size which fits the collision status in current frame when a small

initial frame size is offered and there is large number of tags and hence it will reduce

the efficiency of ILCM. Meanwhile, Schoute algorithm which uses static estimation

has higher system efficiency than ILCM. But, it would also lead to large estimation

error if there is high number of collision tags.

In our proposed approach, Manchester Coding and ideal tag estimation are

used to select optimal frame size. From Figure 4.6, we could notice that the proposed

approach could provide the highest system efficiency and it could remain ~43% even

the number of tags becomes bigger. This is because the proposed approach would

tune the frame size according to the number of collided tag obtained from each

collision slot using Manchester Coding collision bits detection results and thus it

could provide frame size that closely reflected to the collision situation. As a result, it

would able to reduce the number of empty slot and thus the objective to enhance the

system efficiency of RFID system is achieved.

0 200 400 600 800 100036

38

40

42

44

46

48

50

Number of tags

Syste

m E

ffic

iency (

%)

System Efficiency of DFSA

Ideal Tag Esitmation

Schoute

ILCM

Manchester Coding




4.3.2 Comparison between FSA, DFSA and Proposed Approach with Gen2

standard

In this section, we were going to FSA, DFSA and our proposed approach with

Gen2 standard in order to study the slot duration during tag identification process.

There were three scenarios with different BLF values which are 40 kHz, 340 kHz and

640 kHz was designed for the simulations. Different BLF are used is because RFID

system may operate in different frequency bands. For example, RFID system that

operates at low frequency (LF) such as access control, high frequency (HF) is

including payment and tolling system and UHF such as antenna design. Table 4.1

shows the Gen2 parameters used in this project.

Parameter Scenario 1 Scenario 2 Scenario 3

BLF 40 kHz 340 kHz 640 kHz

Tari 16μs 16μs 16μs

Rbl 22μs 22μs 22μs

PRT 276μs 81.88μs 69.75μs

RTCal Tari + 0.75Tari = 28μs Tari + 0.75Tari = 28μs Tari + 0.75Tari = 28μs

TRCal 200μs 23.5294μs 12.5μs

TFS 60μs 60μs 60μs

T1 275μs 31.3529μs 30.8μs

T2 75μs 8.8235μs 4.6875μs

T3 2.5μs 0.2941μs 0.1563μs

M 1 1 1

TRext 4 4 4

TQuery 760μs 565.8824μs 553.75μs

TACK 456μs 456μs 456μs

TQrep 148μs 148μs 148μs

TRN16 675μs 79.4118μs 42.1875μs

TEPC 3475μs 408.8235μs 217.1875μs

TS 5450μs 1174.6μs 934.35μs

TC 1173μs 268.5882μs 225.675μs

TE 425.5μs 180.6471μs 178.9563μs

Table 4.1: Gen2 parameters used in Scenario1, 2 and 3




The rate of tag identification with Gen2 is calculated through the formula as follow:

𝑇𝑖𝑚𝑖𝑛𝑔 𝑜𝑓 𝑒𝑚𝑝𝑡𝑦 𝑠𝑙𝑜𝑡 = 𝑇𝐸 ∗ 𝐸

𝑇𝑖𝑚𝑖𝑛𝑔 𝑜𝑓 𝑐𝑜𝑙𝑙𝑖𝑠𝑖𝑜𝑛 𝑠𝑙𝑜𝑡 = 𝑇𝐶 ∗ 𝐶

𝑇𝑖𝑚𝑖𝑛𝑔 𝑜𝑓𝑠𝑢𝑐𝑐𝑒𝑠𝑠 𝑠𝑙𝑜𝑡 = 𝑇𝑆 ∗ 𝑆

Tag identification rate = 𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑡𝑎𝑔𝑠

((𝑇𝐸∗𝐸)+(𝑇𝐸∗𝐶)+(𝑇𝑆∗𝑆)+𝑇𝑄𝑢𝑒𝑟𝑦)

The parameters showed in Table 4.1 are later applied in FSA with frame size

of 100, 150 and 200, DFSA simulation with ideal tag estimation, Schoute algorithm,

ILCM and our proposed approach, DFSA with Manchester Coding. We started the

experiments from Scenario 1 which has the lowest BLF to Scenario 3 with highest

BLF. Figure 4.7, 4.8 and 4.9 presents the simulation results of FSA, DFSA and

proposed approach in three different scenarios.

Figure 4.7: Scenario 1 tag identification rate of FSA, DFSA and Proposed Approach

0 200 400 600 800 10000

20

40

60

80

100

120

140

Number of tags

Ta

g/s

Tag Identification Rate of FSA with Gen2 standard(Scenario 1)

0 200 400 600 800 1000146

148

150

152

154

156

158

160

Number of tags

Ta

g/s

Tag Identification Rate of DFSA with Gen2 standard(Scenario 1)

L =200

L =150

L =100

Ideal Tag Estimation

Schoute

ILCM

Manchester Coding






As showed in Figure 4.9, we could observe that FSA, DFSA and proposed

approach have the highest tag identification rate in Scenario 3 as compared to

Scenario 1 and 2. This is because when larger BLF is used, the read rate of RFID

system is higher and hence more tags could read within a second.

From Figure 4.7, 4.8 and 4.9, we could also notice that the number of tags/s

that could be recognised by reader in FSA is getting lower when the frame size

became smaller and the number of tags became larger. Especially when L = 100 and

there are 1000 tags, the number of tags could be identified by FSA in three scenarios

are only 3, 14 and 17 per second respectively. This is because the frame size provided

0 200 400 600 800 1000630

640

650

660

670

680

690

700

710

Number of tags

Ta

g/s



Schoute

ILCM

Manchester Coding

0 200 400 600 800 10000

100

200

300

400

500

600

Number of tags

Ta

g/s


L = 200

L = 150

L = 100

0 200 400 600 800 10000

100

200

300

400

500

600

700

Number of tags

Ta

g/s


L = 200

L = 150

L = 100

0 200 400 600 800 1000760

780

800

820

840

860

880

Number of tags

Ta

g/s



Schoute

ILCM

Manchester Coding




did not fit the number of tags to be identified. Consequently, the tags are having

limited number of time slot that is available to be selected and they might select the

same time slot with each other and cause tag collision. As a result, a smaller number

of tags would be successfully recognised by the reader and the reader might need to

spend a longer period of time in order to recognise all the tags. This proved that the

frame size is playing an important role in throughput of FSA as it would affect the tag

identification rate severely if it is selected inappropriately.

For DFSA, the simulation results of ideal tag estimation and Schoute

algorithm tends to provide a stable tag identification rate as compared to FSA which

are 148 tags/s in first scenario, 645 tags/s in second scenario and 784 tag/s in third

scenario. ILCM had the lowest tag identification rate as compared to ideal tag

estimation and Schoute algorithm due to its limitation which could not adjust the

frame size optimally when a small initial frame size is offered. In the traditional

DFSA, it would resolve the tag collision from different collision slots in future read

cycle. Thus, it might have the possibility to cause high rate of tag collision in future

read cycle and more time slots would be used if there is large number of collided tags

found in current read cycle. As a result, the time needed by the collided tags to be

identified by the reader would be longer. Hence, the number of tags that could be

successfully identified by the reader per second in DFSA would be lower as compared

to the proposed approach.

In our proposed approach, if any collision bits are detected from the slot

reservation codes during Manchester Coding collision bits detection process, the

collided tags from each collision slot would store in a collision list. After that, the

reader would resolve the tag collision slot by slot instead of resolving all the tag

collision in one read cycle. Thus, the time slot used in the proposed approach in each

tag redistribution process would be lesser than the traditional DFSA. This would

reduce the number of empty slot and also shorten the duration for the reader to resolve

the collision indirectly. As a result, more tags could be identified successfully by the

reader per second in the proposed approach as compared to the traditional DFSA.

From Figure 4.7, 4.8 and 4.9, we could observe that our proposed approach is able to

provide the highest and stable tag identification rate which are 155 tags/s in first

scenario, 684 tags/s in second scenario and 835 tag/s in third scenario as compared to

FSA and DFSA. Therefore, this achieve the objectives of this project which is to




utilise Manchester Coding to resolve tag collisions during tag identification process

and shorten the timing in Gen2 standard.

In conclusion, the time needed for the reader to successfully read all the tags

in FSA heavily rely on the initial frame size used during tag identification process. If

an inappropriate frame size is selected, it might require a longer time in resolving tag

collision and reduce the system efficiency and tag identification rate of RFID system.

Thus, FSA which use static frame size throughout the tag identification process is not

recommended to be used in resolving tag collision of RFID system anymore. DFSA

which could dynamically adjust the frame size has become a more preferable solution

to resolve the tag collision problem in RFID system as compared to FSA. However,

traditional DFSA would resolve all the tag collisions detected from different time

slots in one read cycle. The problem arises when there is large number of collided tags

found in current read cycle. According to our study, DFSA tends to provide optimal

frame size that fits the current collision situation and this also means that the number

of time slot required for each tag redistribution process would be higher when then

number of collided tag is large. Therefore, the slot duration of DFSA during tag

identification process would be longer when there is large volume of collided tags

found in current read cycle. In our proposed approach, we tried to shorten the slot

duration by resolving the collision slot by slot. This is to reduce the number of

collided tag in each tag redistribution process and further shorten the time needed to

resolve collision. This is because the time slots needed to resolve the collision would

be lower when there is smaller number of collided tags. At the same time, this is also

able to reduce the number of empty slots created in each future read cycle. As a result,

it enables the reader to identify the tags in a shorter duration and also improve system

efficiency of RFID system. In short, the proposed approach is more efficient and

provides shorter slot duration during tag identification process as compared to FSA

and DFSA with ideal tag estimation, Schoute algorithm and ILCM.




CHAPTER 5: CONCLUSION

5.1 Project Review

Some related previous works have been reviewed before developing this

project as we need to have basic the idea of how the main problem which is tag

collision occurred in RFID system and the solutions used to encounter this problem.

In this project we had reviewed the technology involved which is RFID system and

ALOHA-based anti-collision algorithms that help to resolve tag collision. In this

project, we were mainly focus on two algorithms which are FSA and DFSA

respectively.

In this project, we had also provided the critical remark which includes the

strengths and weaknesses for all the reviewed related previous works. This is to

understand the problems encountered in previous works and we could improve or

enhance it in our proposed approach. The main objectives to be achieved in this

project are to mitigate the tag collision problem and shorten the slot duration in RFID

tag identification process.

The proposed approach in our project would adopt DFSA together with

Manchester Coding to reduce the rate of collision in RFID system. Unlike the

conventional DFSA, our proposed approach would allow the tags to reserve their

selected time slot via an 8-bit binary slot reservation code instead of allocating a time

slot directly to a tag. Thus, if the reader found any collision bits in a particular time

slot during Manchester Coding, the time slot would mark as a collision slot and will

go through tag redistribution process to reserve a new time slot. Besides, the proposed

approach would resolve the tag collision slot by slot. Therefore, it could lower the

number of time slots used and shorten the time needed by the reader to resolve the

collision as the number of collided tags involved in each future read cycle is reduced.

As a result, it enables the reader to recognise more tags during tag identification

process and enhance the system efficiency of RFID system.




5.2 Discussion

The aim of this project is to pinpoint the importance of anti-collision algorithm

in RFID system. As we know, tag collision is always the biggest obstacle for RFID

system to achieve high identification accuracy and tag identification rate in RFID

system. Thus, it is important to study and improve the existing related works in order

to further mitigate the tag collision problem in RFID system.

In this project, we have proposed a new timing concept which applying a bit-

tracking technology, Manchester Coding in DFSA. Before that, we had done the study

for both FSA and DFSA which are ALOHA-based anti-collision algorithms. This is

because these two algorithms are widely used in resolving this problem. In this paper,

we had found that FSA is very time consuming while resolving tag collision due to its

fixed frame size. Thus, we had adopted DFSA which could dynamically adjust the

frame size based on the slot information in our proposed approach.

Furthermore, we also recognised a significant limitation of DFSA which is

resolving the collision of all the collided tags in one read cycle. This would increase

the number of time slot and time needed for the reader to resolve the collision when

there is a large number of a collided tag found in current read cycle. Consequently,

this would increase the time needed during the tag identification process. Thus, we

tried to resolve this issue by letting the tag to reserve their time slots using an 8-bit

reservation code and resolve the collision in slot basis after detecting collision bits

using Manchester Coding from each time slot. According to our simulation results, we

found that the number of slot used in our proposed approach is lower than FSA and

DFSA as the number of collided tags involved in future read cycle is reduced. Besides,

it is also able to provide a stable and higher tag identification rate and system

efficiency which is ~43% during tag identification process. This is because the time

needed for reader to resolve the tag collision would be reduced if the number of

collided tag becomes smaller and therefore more tags could be recognised each read

cycle. Thus, this achieved the main objectives of this project which is to utilise

Manchester Coding to resolve tag collisions during tag identification process and

shorten the timing in Gen2 standard.




5.3 Contributions

During the implementation and development of our proposed approach, we

had recognised the limitations of existing ALOHA-based anti-collision algorithms.

For instance, FSA which remains its initial frame size throughout the tag

identification process would cause high rate of tag collision when there is large

volume of tag is involved. For DFSA, the frame size is always heavily relies on the

tag estimation result and thus the adjusted frame size might not be optimal when there

is any estimation error. Besides, DFSA would resolve all the collided tags determined

in current read cycle with a new frame size in a new read cycle. Consequently, the

reader might spend longer time and use higher number of time slot to resolve the

collision if there is large volume of collided tag found in current read cycle.

In our proposed approach, we would detect collision through Manchester

Coding by tracing the slot reservation codes in each time slot. If there is any collision

bits detected, the collided tags in a time slot would need to restart the whole slot

reservation process in future read cycle with a new frame size. Unlike the traditional

DFSA, our proposed approach would solve the collision slot by slot. This could cut

down the number of collided tags and time slot used in each future read cycle.

Therefore, the success rate for a tag to be recognised by the reader would be higher as

the reader might spend a shorter period of time to resolve the collision. As a result,

our proposed approach could help to reduce the slot timing used in tag identification

process and improve the system efficiency of RFID system.

In a nutshell, the proposed approach of this project could mitigate the tag

collision problem and enhance identification time in RFID system. Therefore, it

enables RFID system to achieve its main advantage which is fast data reading and

helps to improve the productivity of the industries or application systems which

employed RFID system in their business operations.




5.4 Future works

During the implementation of the proposed approach of this project, we found

that the time needed to complete the simulation for 1000 experiments would last for

hours or days while the number of tags involved is becoming larger. This happened is

due to there might be collision occurred during the tag redistribution process. Thus,

these tags might go through many rounds of tag redistribution process if there is any

collision detected.

Besides, the slot reservation code used in the slot reservation code

regeneration process in proposed approach would only cater up to maximum 255 tags

which mean that for each collision slot the maximum number of tags that could

receive a new reservation code is only up to 255. This is because the maximum

number of tags we used in our experiments is only up to 1000 tags and thus we used

randperm() function to regenerate unique slot reservation code for the tags which used

the same code to reserve a time slot in slot reservation code regeneration process. As

we know, the maximum number for 8-bit binary string is 255. Thus, the proposed

approach might not able to cater the case of a collision slot which contained 255 tags

and with duplicate slot reservation code.

Thus, the future enhancement for our proposed method should have reduced

the chances of collision during tag redistribution process. This is to further reduce the

time needed for the reader to resolve collision and the number of time slot used in

each future read cycle. Besides, the future work should provide a way to cater for the

cases when there are more than 255 tags with duplicate slot reservation code as there

might more than 1000 tags to be recognised in real life application.

Bibliography



BIBLIOGRAPHY

Bang, O., Kim, S. and Lee, H. (2009) ‘Identification of RFID Tags in Dynamic

Framed Slotted ALOHA’, Advanced Communication Technology, 01, pp. 354–

357.

Chen, Y., Su, J. and Yi, W. (2017) ‘An efficient and easy-to-implement tag

identification algorithm for UHF RFID systems’, IEEE Communications Letters, ,

pp. 1–1. doi: 10.1109/lcomm.2017.2649490.

Cheng, T. and Jin, L. (2007) ‘Analysis and simulation of RFID anti-collision

Algorithms’, The 9th International Conference on Advanced Communication

Technology,. doi: 10.1109/icact.2007.358450.

content (2008) Using RFID & IPv6. Available at:

http://ipv6.com/articles/applications/Using-RFID-and-IPv6.htm (Accessed: 18

August 2016).

Use, I.T. of, Policy, P. and Trademarks (2002) The different types of RFID systems.

Available at: http://www.impinj.com/resources/about-rfid/the-different-types-of-

rfid-systems/ (Accessed: 15 August 2016).

Jin, X., Wei, D., Xu, Y., Jin, L. and Huang, X. (2015) ‘A novel RFID tag estimation

algorithm based on DFSA’, 2015 IEEE 5th International Conference on

Electronics Information and Emergency Communication, . doi:

10.1109/iceiec.2015.7284479.

Kaewsirisin, S., Supanakoon, P., Promwong, S., Sukutamtanti, N. and Ketprom, U.

(2008) ‘Performance study of dynamic framed slotted ALOHA for RFID

systems’, 2008 5th International Conference on Electrical

Engineering/Electronics, Computer, Telecommunications and Information

Technology, . doi: 10.1109/ecticon.2008.4600459.

Bibliography



Landaluce, H., Perallos, A. and Angulo, I. (2014) ‘Managing the number of tags bits

transmitted in a bit-tracking RFID collision resolution protocol’, Sensors, 14(1),

pp. 1010–1027. doi: 10.3390/s140101010.

McDowell, G. (2009) How does RFID technology work? [Technology Explained].

Available at: http://www.makeuseof.com/tag/technology-explained-how-do-rfid-

tags-work/ (Accessed: 22 June 2016).

Qing, Y., Jian-cheng, L., Hong-yi, W. and Rong-jun, S. (2012) ‘A dynamic framed

slotted ALOHA anti-collision algorithm based on tag-grouping for RFID

systems’, 2012 IEEE 11th International Conference on Solid-State and

Integrated Circuit Technology, . doi: 10.1109/icsict.2012.6467907.

Nov-2013, version 2.0 EPC™ radio-frequency identity protocols generation-2 UHF

RFID specification for RFID air interface protocol for communications at 860

MHz – 960 MHz version 2.0.0 ratified (2013) Available at:

http://www.gs1.org/sites/default/files/docs/epc/uhfc1g2_2_0_0_standard_201311

01.pdf (Accessed: 20 August 2016).

Raja, T. and Perumal, V. (no date) T RAJA AND V PERUMAL: ADVANCED

DYNAMIC FRAMED SLOTTED ALOHA ADVANCED DYNAMIC

FRAMED SLOTTED ALOHA. Available at:

http://ictactjournals.in/paper/IJCT_Vol3_Iss2_P5_548_550.pdf (Accessed: 12

August 2016).

Shakiba, M., Zavvari, A. and Sundararajan, E. (2011) ‘Fitted dynamic framed slotted

ALOHA anti-collision algorithm in RFID systems’, ICIMU 2011 : Proceedings

of the 5th international Conference on Information Technology & Multimedia, .

doi: 10.1109/icimu.2011.6122776.

Solic, P., Radic, J., Dujmic, H., Saric, M., Russo, M., Stella, M., Begusic, D. and

Rozic, N. (2013) ‘Improved Linearized Combinatorial model (ILCM) for optimal

frame size selection in ALOHA-based RFID systems’, 2013 IEEE International

Bibliography



Conference on Communications Workshops (ICC), . doi:

10.1109/iccw.2013.6649396.

Soli´c, P. and Stella, M. (2013) ‘Tags/s -RFID reader-tag communication throughput

using gen2 Q-algorithm frame adaptation scheme’, Recent Advances in

Information Science, pp. 100-105

Su, J., Sheng, Z., Hong, D. and Wen, G. (2016) ‘An effective frame breaking policy

for dynamic framed slotted Aloha in RFID’, IEEE Communications Letters, 20(4),

pp. 692–695. doi: 10.1109/lcomm.2016.2521839.

Tong, Q., Zou, X. and Tong, H. (2009) ‘Dynamic framed slotted ALOHA algorithm

based on Bayesian estimation in RFID system’, 2009 WRI World Congress on

Computer Science and Information Engineering, . doi: 10.1109/csie.2009.248.

Appendices

A-1

APPENDIX A - Weekly Report

FINAL YEAR PROJECT WEEKLY REPORT

(Project II)

Trimester, Year: Year 4 Semester 1 Study week no.: Week 2

Student Name & ID: Lee Khai Yi, 14ACB00454

Supervisor: Dr Robithoh Annur

Project Title: ALOHA-Based Radio-Frequency Identification (RFID) System With

Early Frame Adjustment

1. WORK DONE

Find work related articles and figure out the ideas on how to apply Manchester

Coding in DFSA.

2. WORK TO BE DONE

Design the logic flow and flow diagram of proposed approach

The possible limitations of the proposed method

3. PROBLEMS ENCOUNTER

Since Manchester Coding technology is always applied on Binary Tree algorithm

and thus applying it in DFSA is a new idea.

4. SELF EVALUATION OF THE PROGRESS

Understand the basic idea of how Manchester Coding work in Binary Tree

algorithm

Find out the rough idea of how to apply Manchester Coding in DFSA

_________________________ _________________________

Supervisor’s signature Student’s signature

Appendices

A-2


(Project II)






1. WORK DONE

Literature review of existing improved RFID anti-collision algorithms

Understand how Manchester Coding works

2. WORK TO BE DONE

Produce flow diagram for proposed idea

Program the proposed idea


Some proposed idea might not be implemented and thus other alternative solutions

need to be figured out


Understand the program flow of proposed idea and bit tracing technology of

Manchester Coding

_________________________ _________________________


Appendices

A-3


(Project II)






1. WORK DONE

Bit-tracking of Manchester Coding

Calculate the tag identification rate and system efficiency of proposed

approach

2. WORK TO BE DONE

Apply Gen2 timing in proposed method

Check logic flow of proposed method


The time taken for proposed method to identify the tags is extremely long and the

binary code generated could only cater for not more than 10000 tags.


Done Manchester code generation for collision checking purpose and slot selection

method used in the proposed method.

_________________________ _________________________


Appendices

A-4


(Project II)






1. WORK DONE

Program code for proposed approach

Simulations for proposed approach

2. WORK TO BE DONE

Resolve the slot reservation code collision occurred when there are more than

one tags using the same slot reservation code to reserve a time slot

Redesign the checking process for collision slots


The tags that reserved the same time slot were not inserted into remaining tag

list.

The unread tags which has the same bit pattern as the tags stored in the success

list are removed from the list.


Change the slot reservation code generation using randi() instead of

randperm().

Change the checking process for collision time slot and redesign the slot

reservation process when there is a collision detected by using Manchester

Coding.

_________________________ _________________________


Appendices

A-5


(Project II)






1. WORK DONE

Report writing for Chapter 1, 2 and 3

2. WORK TO BE DONE

Obtain the simulation results of proposed approach

Graph plotting for proposed approach and existing ALOHA-based anti-collision

algorithms

FYP poster design


N/A.


End of program enhancement

_________________________ _________________________


Appendices

A-6


(Project II)






1. WORK DONE

Report and poster design enhancement

2. WORK TO BE DONE

Enhance Chapter 3 discussion

Perform Turnitin check


N/A.


Reorganise some contents and finalise the report

_________________________ _________________________


Appendices

B-1

APPENDIX B - Poster

Appendices

C-1

APPENDIX C - Turnitin Similarity Report

Appendices

C-2

Appendices

C-3

Appendices

C-4

Appendices

C-5

Appendices

C-6

FACULTY OF INFORMATION AND COMMUNICATION

TECHNOLOGY

Full Name(s) of Candidate(s)

Lee Khai Yi

ID Number(s)

14ACB00454

Programme / Course Bachelor of Information Technology (Hons) Communications and

Networking

Title of Final Year Project ALOHA-Based Radio-Frequency Identification (RFID) System With


Similarity Supervisor’s Comments (Compulsory if parameters of originality exceeds the limits approved by UTAR)

Overall similarity index: ___ %

Similarity by source Internet Sources: _______________% Publications: _________ % Student Papers: _________ %

Number of individual sources listed of more than 3% similarity:

Parameters of originality required and limits approved by UTAR are as Follows:

(i) Overall similarity index is 20% and below, and (ii) Matching of individual sources listed must be less than 3% each, and (iii) Matching texts in continuous block must not exceed 8 words

Note: Parameters (i) – (ii) shall exclude quotes, bibliography and text matches which are less than 8 words.

Note Supervisor/Candidate(s) is/are required to provide softcopy of full set of the originality

report to Faculty/Institute

Based on the above results, I hereby declare that I am satisfied with the originality of the

Final Year Project Report submitted by my student(s) as named above.

______________________________ ______________________________ Signature of Supervisor Signature of Co-Supervisor

Name: __________________________

Name: __________________________

Date: ___________________________ Date: ___________________________

Universiti Tunku Abdul Rahman

Form Title : Supervisor’s Comments on Originality Report Generated by Turnitin

for Submission of Final Year Project Report (for Undergraduate Programmes)

Form Number: FM-IAD-005 Rev No.: 0 Effective Date: 01/10/2013 Page No.: 1of 1

REPORT STATUS DECLARATION FORMeprints.utar.edu.my/2466/1/CN-2017-1400454.pdf · Title: Aloha-Based Radio-Frequency Identification (RFID) System With Early Frame Adjustment Academic

Documents